Seven models, available commercially for the self-measurement of blood pressure, were subjected to a validation procedure in which three devices of each model were tested by observers who were trained to a high standard of accuracy. The models were the Omron HEM-400C, the Philips HP5308, the Healthcheck 'Cuffless' CX-5 060020, the Nissei Analogue Monitor, the Philips HP5306/B, the Systema Dr MI-150 and the Fortec Dr MI-100. The validation programme had a number of unique features which included assessment of interdevice variability before and after 1 month of home use, and a new form of analysis, which we term 'clinical', based on the likely influence of three grades of device inaccuracy on patient management. In the main validation phase, one device of each model was compared with simultaneous measurements made by two 'blinded' observers using a standard mercury sphygmomanometer (PyMaH Corporation, New Jersey, USA) in the same arm in 85 subjects with a wide range of blood pressures. Three models (the Healthcheck 'Cuffless' CX-5 060020, the Systema Dr MI-150 and the Fortec Dr MI-100) failed the interdevice variability tests and did not reach the main validation test. Two models (the Omron HEM-400C and the Philips HP5306/B) failed on the criteria set down by the American National Standard for Electronic or Automated Sphygmomanometers, as well as the 'clinical' criteria. The remaining two models (the Nissei Analogue Monitor and the Philips HP5308) were acceptable for the measurement of systolic blood pressure by both methods of analysis but failed in the 'clinical' analysis for diastolic blood pressure. The mercury sphygmomanometer was comfortably within the criteria for both methods of analysis.